Struct RuntimeHandle 

pub struct RuntimeHandle { /* private fields */ }

A cheap clone of the runtime, safe to hand to flavors and connection tasks.

Implementations

impl RuntimeHandle

pub fn ids(&self) -> &SharedIds

The runtime-wide identifier allocator.

Public because the gateway needs a connection’s session before any shard has rendered it: ServerSync carries it, and the handshake happens before the first turn.

pub fn peers(&self) -> &Arc<Peers>

pub fn next_connection(&self) -> ConnectionId

Reserve the next connection identifier. Never reused.

pub fn session_for( &self, connection: ConnectionId, ) -> Result<SessionId, Exhausted>

The session a connection will be rendered under, allocated now so the handshake can announce it.

Errors

Once the session space is spent.

pub fn create_shard<L: ShardLogic>( &self, build: impl FnOnce(ShardHandle) -> L, ) -> ShardHandle

Create a shard and start its task.

The chicken and the egg - a flavor usually wants to wake the shard that owns it - are resolved by a closure: the handle exists before the logic that will hold it.

pub fn destroy_shard(&self, shard: ShardId, reason: &str)

Destroy a shard.

The guide leaves the policy for its connections open (17.4). The safe default, and what this does, is to close them: a connection left pointing at a shard that no longer renders would hold a view nothing can ever update. A flavor that wants a fallback shard migrates them first and destroys afterwards.

pub fn move_connection(&self, connection: ConnectionId, to: ShardId)

Move a connection to another shard.

Not a disconnect followed by a connect. The source shard hands over the view the client still holds, and the destination plans one transition from it onto its own tree, which is the same slow path a scope change takes: whatever the two trees have in common does not flicker.

Doing it the other way round is not merely wasteful, it disconnects the official client. It keeps itself in its own model after a UserRemove naming its own session, so the ChannelRemove that follows looks to it like the server removing an occupied channel, which it treats as a protocol violation.

REF: references/mumble/src/mumble/Messages.cpp : msgUserRemove calls removeUser only if (pDst != pSelf); msgChannelRemove logs “Protocol violation. Server sent remove for occupied channel.” and disconnects when UserModel::removeChannel(c, true) refuses.

pub fn attach( &self, peer: &Arc<Peer>, shard: ShardId, ) -> Result<Receiver<()>, AttachError>

Attach a connection that has just been routed.

Returns the receiver that fires once the shard has accepted the connection’s first transition, which is what the handshake waits on before sending ServerSync.

Errors

When the shard does not exist, or its mailbox is full.

pub fn detach(&self, connection: ConnectionId, shard: ShardId, reason: &str)

Tell a connection’s shard that it is gone.

pub fn send( &self, shard: ShardId, command: ShardCommand, ) -> Result<(), AttachError>

Forward a command to a shard.

Errors

When the shard does not exist or its mailbox is full.

pub fn status(&self) -> Vec<ShardStatus>

What an operator would want on a status page.

pub fn shard_handle(&self, shard: ShardId) -> Option<ShardHandle>

Trait Implementations

impl Clone for RuntimeHandle

fn clone(&self) -> RuntimeHandle

Returns a duplicate of the value.

fn clone_from(&mut self, source: &Self)

Performs copy-assignment from source.

impl Debug for RuntimeHandle

fn fmt(&self, formatter: &mut Formatter<'_>) -> Result

Formats the value using the given formatter.